Having problems with screen grid current.

[Bastard Circuits]

Here's my problem. I just built an amp with 250V on the plates of KT66. 6.6K output transformer primary. Fixed bias with bias set at 100ma current draw. Here's the thing. Everything is fine until signal is applied. All of a sudden, plate current drops to practically nothing (measured at the plate), and the screen grids start drawing massive current (20-60ma depending on bias level). I've got about 240V initially on the screens with 1K 5W wirewounds. Choke in the power supply. What's going on here?

[Ron]

The KT66 is turning on very hard, like a switch. The plate voltage goes to zero probably because of the voltage drop across the plate load when the plate current is high. Most of the B+ is across the load, not the tube.

I would look at (1) the amplitude and shape of the input signal to the grid with a scope, (2) make sure you have a plate load that's in the ball park, and (3) check your power supply to see whether it's totally wilting or reasonably constant.

[skyboltone]

I don't think you can operate KT66s at 250VDC and 100ma zero signal plate current. The book says more like 162ma bias current. Even then you only have -17.5v on the grids. Most guitar pre amp sections will swing a bigger voltage than that. When you strum the guitar you are swinging the grids positive with respect to the cathode. Run away tubes result. With the screens running at 240VDC they just become secondary plates in this scenario. You need a redesign.

My .02

Dan

[Bastard Circuits]

Ron, just wanted to clarify, the plate is not dropping voltage, its dropping current. The power supply seems to be very stable. What are my options on making this voltage/current scenario work? Preamp attenuation?

[Ron]

Thanks for the correction, it makes a difference. I agree with skyboltone, check your input signal and start with a small amplitude. If you get amplification without the plate and screen current doing odd things, then you can increase the signal until they do. Who knows what the tube will do when you have a huge signal input compared to what it can linearly ampify.

[Bastard Circuits]

Well, i had been thinking of going with lowering the value of the plate resistors and grid reference resistors in the phase inverter to 33k ala later Twin Reverb. Just haven't tried it yet. This should curb the size of the waveform, no?

[Ron]

I don't know what your setup is, so it's hard to comment on point.

If you have a signal generator with an attenuator, then use it.

If you're driving the amp with a guitar, man, you have a problem controlling amplitudes!!

Somewhere you should be able to tack solder a pot into the signal path and use it as a voltage divider/attenuator. Start at zero and crank it slowly.

Is this a single-ended or push-pull type amp?

[Bastard Circuits]

Marshall / Bassman style preamp - push-pull output stage. Pretty standard design except for the Class A output stage at 250V with fixed bias and no negative feedback. The basic premise of this design is to get output distortion at lower volume.

[skyboltone]

Put a PI network with 750 ohm parallel legs and a 2.9K series leg between the output of the preamp and the PI. This will lose 20db of signal to allow you to continue to tweek. If it's not enough try raising the series and dropping the parallel elements until it will amplify.

This ain't gonna sound good but you can play.

If it still runs away try 600 ohm parallel legs and a 10k series leg.

That's 50db

[Ron]

You can use a voltage divider (1M ohm "volume pot") at the input to the phase inverter to control your signal level for testing. Start at zero signal and listen to your guitar, if you don't have a sig gen and scope.

[Bastard Circuits]

I tried the preamp attenuation thing with no result. I should clarify that the amp works. Play guitar through it it sounds really good up to maybe 10 o'clock on the volume control. This point coincides with the screen grids drawing massive current and the output (on the scope) totally cliping hard, and now i've noticed the control grids are drawing current as well. Starting to suspect the bias supply. Tried changing the Marshall / Bassman style supply to a Fender style, no luck. Tommorow I'll try cathode biasing it. Anyone tried building a guitar amp with plate voltages this low and fixed bias supply? All of the class a'ish amps i'm familiar with are cathode biased. Someone said somewhere that most class a amps were built cathode biased because of the added expense of a bias supply, but at so low of a supply voltage and high current demand, is a 50V bias tap messing with things? I just can't think of anything to really compare it to.

[tictac]

sounds like you've got some parasitic oscillation problems. Can be a bear to find out the cause. Chassis layout (esp. transformer location), lead dress, and grounding system are crucial.

[Bastard Circuits]

Yeah, it's not parasitic. Apparently, the plate to plate impedance for goes down when the plate voltage is that low. This would explain why a pair of 6V6's worked better (higher plate to plate to start with).